kaffo / coursework Goto Github PK

Outputs of association rule mining can be sent to clients, from backend, at key events (e.g. install of package …)

We could use IP address, or some crypto hash thingy to uniquely identify machines -
When you first install the plugin, do you get a unique id for that client? Issues with local reinstalls, etc.

when user does dnf install, client sends message to backend (well-known web service), and backend records this install event in its database

There are lots of options here - we could go for something very traditional like MySQL etc.
Or we could go for a scalable solution like Mongo DB / etc.
Or we could go for a trivial sqlite or backing plain txt csv file.
These are design decisions you will need to comment on in your dissertation :-)

get dnf config file reading working

implement association rule mining for packages in backend. Could be done with database or directly in Python - explore the options before implementing whichever seems most appropriate